This is another way to and its a little more dynamic, and saves the

DECLARE @CustOrder TABLE

(

CustId Int

,SalesValue Money

)

INSERT INTO @CustOrder

(CustId,SalesValue)

Values

(100813, 124)

,(100813, 110)

,(100794, 297)

,(100794, 99)

,(100794, 101);

WITH CTE_data

As

(

SELECT ROW_NUMBER() OVER (PARTITION BY CustId Order by CustId) rn

,CustId

,SalesValue

From @CustOrder

)

,CTE

AS

(

SELECT

CustId

,Max(CASE rn WHEN 1 THEN 1 ELse 1 END) Number1

,Max(CASE rn WHEN 1 THEN SalesValue ELse 0 END) Salesvalue1

,Max(CASE rn WHEN 2 THEN 2 ELse 2 END) Number2

,Max(CASE rn WHEN 2 THEN SalesValue ELse 0 END) Salesvalue2

,Max(CASE rn WHEN 3 THEN 3 ELse 3 END) Number3

,Max(CASE rn WHEN 3 THEN SalesValue ELse 0 END) Salesvalue3

From CTE_Data

Group by CustId

)

SELECT

Number,CustId, SalesValue

FROM

CTE

CROSS APPLY

(VALUES

(Number1,SalesValue1)

,(Number2,SalesValue2)

,(Number3,SalesValue3)

) x (Number,SalesValue)

Not sure of the performance on a very large table but splitting out the first CTE that adds a row number into a Temporary table.